Serologic characteristics of ceftriaxone antibodies in 25 patients with druginduced immune hemolytic anemia

Patricia A. Arndt, Regina M. Leger, and George Garratty

BACKGROUND: Ceftriaxone, a third-generation cephalosporin, is commonly used
to prevent and treat infections. Since 1987, it has been the second most common
cause of drug-induced immune hemolytic anemia (DIIHA) investigated in our
laboratory.
STUDY DESIGN AND METHODS: Samples from 79 patients (1987-2010),
suspected of having DIIHA caused by ceftriaxone, were studied for the presence of
ceftriaxone antibodies. Direct antiglobulin tests (DATs) and tests with ceftriaxone
treated red blood cells (RBCs) or untreated and enzyme-treated RBCs in the
presence of ceftriaxone were performed.
RESULTS: Twenty-five (32%) of the 79 patients had antibodies to ceftriaxone
detected. Seventeen (68%) of the 25 patients were children; reactions in children
were usually dramatic and severe. Nine (36%) of the 25 patients had fatal DIIHA.
Nineteen of the 25 samples had DATs performed by our laboratory; 100% of
samples were reactive with anti-C3 and 47% were reactive with anti-IgG. All 25
sera had ceftriaxone antibodies detected when testing untreated or ficin-treated
RBCs in the presence of ceftriaxone (resulting in agglutination, hemolysis or
sensitization of test RBCs). These antibodies were primarily IgM and reactivity was
enhanced by testing ficin-treated RBCs. Sixteen (64%) of the 25 sera reacted with
test RBCs when no ceftriaxone was added in vitro; this was most likely due to the
transient presence of drug or drug-immune complexes in the patients circulation at
the time that the blood samples were drawn.
CONCLUSION: Ceftriaxone antibodies can cause severe intravascular hemolysis.
Complement can usually be detected on the patients RBCs and IgM antibodies
are usually detected in the patients serum. Ceftriaxone, a third-generation
cephalosporin, has been in use since 1982. A generic form of this broad-spectrum
antibiotic became available in 2005. Ceftriaxone is administered parenterally once

or twice daily (the half-life is approx. 6-8 hr) and is eliminated unchanged in the
urine and also in bile.1 Drug-induced immune hemolytic anemia (DIIHA) is rare
with an estimated incidence of about one per million of the population.2 The first
case of DIIHA due to ceftriaxone was detected in our laboratory in late 1987,
reported at the AABB meeting in 1988, and published in 1991.3 The patient was a
52-year-old woman who received ceftriaxone for an abdominal infection. Not only
was this the first case of DIIHA associated with ceftriaxone, but it was also the first
case of fatal cephalosporin-induced hemolytic anemia and the second case in
which a cephalosporin antibody showed in vivo and in vitro characteristics
associated with the so-called immune complex mechanism (e.g., intravascular
hemolysis in vivo, reactions with untreated red blood cells [RBCs] in the presence
of a solution of drug in vitro). Before this, the few reported cephalosporin antibodies
associated with DIIHA had shown characteristics similar to penicillin antibodies
(e.g., extravascular hemolysis in vivo, reactions with drugtreated RBCs in vitro).
Since this first reported case, there have been at least 27 further single case
reports of DIIHA due to ceftriaxone with serologic evidence to support the
diagnosis.4-29 Sixteen (57%) of these 28 patients with DIIHA due to ceftriaxone
were children and 9 of the 28 patients died (six children, three adults).3,58,10,19,23,28 It should be noted that the dramatic cases of DIIHA are more likely
to be recognized and published. In all of these published cases, the antibody to
ceftriaxone was detected when reagent RBCs were tested in the presence of a
solution of ceftriaxone; RBCs treated with ceftriaxone either were nonreactive or
were not tested. In three cases, ceftriaxone antibodies were detected only in the
presence of ex vivo drug (urine from a person taking ceftriaxone)
ABBREVIATION: DIIHA = drug-induced immune hemolytic anemia.
From the American Red Cross Blood Services, Southern California Region,
Pomona, California. Address correspondence to: Patricia A. Arndt, MS, MT(ASCP)
SBB, American Red Cross, 100 Red Cross Circle, Pomona, CA 91768; e-mail:
arndtp@usa.redcross.org. Received for publication May 17, 2011; revision
received July 11, 2011; and accepted July 12, 2011. doi: 10.1111/j.15372995.2011.03321.x TRANSFUSION 2012;52:602-612.

Our research laboratory specializes in immune hemolytic anemias30 and we are

referred cases of suspected DIIHA from a variety of hospitals and reference
laboratories (primarily, but not exclusively, in the United States). In the 30-year
period from 1979 to 2008, we identified 150 cases of DIIHA;31 from 2009 to 2010
we detected 19 more. Fifty-one percent of the total 169 DIIHA cases were due to
cefotetan; the second most common cause was ceftriaxone (15%). In the past 5
years we have seen a change; cefotetan has dropped to the third most common
cause of DIIHA (eight cases from 2006 to 2010), ceftriaxone remains as the
second most common cause (11 cases from 2006 to 2010), and piperacillin has
risen to currently be the most common cause of DIIHA (17 cases from 2006 to
2010) that we see in our laboratory. This report gives details of the 25 cases of
DIIHA due to ceftriaxone antibodies we encountered since 1987.
MATERIALS AND METHODS
From November 1987 to the end of 2010, we were sent samples from a total of 79
patients suspected of having DIIHA due to ceftriaxone. Cases were screened via
telephone in an attempt to exclude those that did not fit posible DIIHA, for example,
the patient was not hemolyzing, the hemolysis was not temporally related to the
administration of drug, or the direct antiglobulin test (DAT) was negative at the time
of the hemolytic anemia. The testing protocol included performing aDAT on the
patients RBCs and testing the patients serum (and, in the early cases, an eluate
prepared from the patients RBCs) for the presence of antibodies to ceftriaxone.
Initially, testing for the presence of ceftriaxone antibodies was performed by two
methods: 1) with ceftriaxone-treated RBCs and 2) with untreated or enzymetreated RBCs in the presence of a solution of ceftriaxone. In 1999 we discontinued
routine testing of eluates and ceftriaxone-treated RBCs.
DATs
DATs were performed by the test tube method. The patients washed
ethylenediaminetetraacetate (EDTA) RBCs were tested with anti-IgG (American

Red Cross, Rockville, MD; or Ortho-Clinical Diagnostics, Raritan, NJ), anti-C3 (inhouse), anti-IgM (in house or CLB/Sanquin, Amsterdam, Netherlands), anti-IgA
(Atlantic Antibodies, Westbrook, ME; orTago, Burlingame, CA), and a 6%bovine
albumin control. The in-house anti-C3 and anti-IgM reagents were prepared by
injecting rabbits with purified proteins. The anti-C3, anti-IgM, and anti-IgA reagents
were diluted in 6% bovine albumin and standardized for use by the test tube
method as previously described
Testing sera and eluates for ceftriaxone antibodies
Ceftriaxone powder was supplied by the submitting institutions as Rocephin
(Hoffman-LaRoche, Nutley, NJ) or generic ceftriaxone (Sandoz GmbH, Kundl,
Austria). The solubility of ceftriaxone in water at 25C is approximately 400
mg/mL32 so there were no difficulties preparing solutions of ceftriaxone in a waterbased buffer such as phosphate-buffered saline (PBS). Ficin, used to prepare
enzyme-treated RBCs, was obtained from Sigma-Aldrich (St Louis, MO). Eluates
were prepared from patients RBCs by either a xylene method33 or a commercial
acid elution kit (Amtec Diagnostics International, Conroe, TX; or Gamma
Biologicals, Houston, TX). Starting in 1995, the wash solution used to prepare
RBCs for acid elution was modified.34
Drug-treated RBCs
Group O donor RBCs were incubated with ceftriaxone in attempts to prepare drugcoated RBCs for testing. Methods included 1) incubating 40 mg/mL solutions of
ceftriaxone prepared in either PBS (pH 7.0-7.4) or 0.1 mol/L sodium barbital buffer
(Sigma Chemical Co., St Louis, MO; pH 9.8) with 1/10 volume RBCs for 1 to 2
hours at 37C or 2) a carbodiimide coupling method.30,35 For the carbodiimide
method,

freshly

prepared

solution

of

1-ethyl-3-(3-dimethyl-aminopropyl)

carbodiimide hydrochloride (Sigma) was added to a mixture of RBCs and drug in

PBS (6.7 mg/mL) and incubated for 50 minutes at 4C. The RBCs in the drugcarbodiimide mixture were added to cold 2% EDTA-PBS (Sigma; pH 7.2) and then
washed. Patients ser were tested with drug-treated and untreated RBCs by a

previously described test tuve method.30 Briefly, two drops of patients serum were
incubated with one drop of treated or control untreated RBCs. After 1 hour of
incubation at 37C, tests were examined for hemolysis and agglutination.
Antiglobulin tests were performed using anti-IgG,-C3d (Immucor, Norcross, GA; or
Ortho). As negative controls, pools of normal sera or plasma were tested in parallel
with the patients sera. In a few cases, washes for the antiglobulin testwere
performed using PBS containing ceftriaxone (0.01 mg/mL).
Untreated or enzyme-treated RBCs in the presence of drug solution
Patients sera were also tested, with and without the addition of pooled fresh
normal sera as a source of comple ment, in the presence of a solution of
ceftriaxone with untreated and ficin-treated RBCs by a tube method as previously
described.30 Briefly, two drops of patients serum with or without two drops of fresh
normal ser were incubated with two drops of ceftriaxone (1 mg/mL in PBS) plus
one drop of either untreated or ficin-treated group O reagent RBCs for 1 to 2 hours
at 37C. If a patient had alloantibodies, then reagent RBCs negative for the
appropriate antigens were used. Tests were examined for hemolysis or
agglutination and then an antiglobulin test was performed using anti-IgG,-C3d. If
agglutination was noted after the 37C incubation, tests were examined for
carryover of agglutination after the washes for the antiglobulin test but before the
addition of the antiglobulin sera. Parallel tests with PBS added to patients sera
instead of ceftriaxone, or with fresh normal sera instead of patients sera, served as
controls.
Additional studies
Sera from 22 patients with ceftriaxone antibodies had titrations performed by the
tube method. Doubling dilutions of serawere prepared either in PBS (for
agglutination and antiglobulin test titers) or in pooled fresh normal sera (for
hemolysin titers) and tested with untreated or ficin-treated reagent RBCs in the
presence of 1 mg/mL ceftriaxone. Sera from 11 patients were tested using a

modification of a previously described gel method.36 Two different types of gel

cards were used: 1) buffered gel cards (Micro Typing Systems, Pompano Beach,
FL) to detect direct agglutination and 2) anti-IgG,-C3d gel cards (MicroTyping
Systems). Briefly, 50 mL of
untreated

or

ficin-treated

RBCs (diluted in PBS to 1%

vol/vol) was incubated with 25
mL of patients serum with or
without 25 mL of pooled fresh
normal

sera

source)

(complement

plus

25

mL

ceftriaxone (1 mg/mL) or PBS

in both types of gel cards for
60 minutes at 37C and then
centrifuged. Controls of 25
mL of pooled fresh normal
sera

(complement

source)

plus 25 mL of ceftriaxone (1
mg/mL) or PBS were tested in
parallel.

To

determine

immunoglobulin classes, sera

from

15

patients

with

antibodies to ceftriaxone were

treated with the sulfhydryl reagents, dithiothreitol (DTT) or 2-mercaptoethanol, as
previously described,37,38 and then tested by the tube method with untreated
reagent RBCs in the presence of ceftriaxone.
RESULTS
Patients with antibodies to ceftriaxone Antibodies to ceftriaxone were detected
in sera from 25 (32%) of the 79 patients suspected of DIIHA caused by ceftriaxone.
Three patients with ceftriaxone antibodies were identified in the 1987 to 1990

period, six patients from 1991 to 2000, and 16 patients from 2001 to 2010.
Fatalities occurred in 6 of 17 (35%) children and three of eight (38%) adults. Table
1 shows some characteristics of the 25 patients and their hemolytic anemias. Eight
of these cases (two adults, six children) have been published as single case
reports.3,6,7,13-15,26,27 The serologic results of the first eight patients with DIIHA
due to ceftriaxone were summarized in 1999.39 The antibody from the first patient3
was used to study serologic cross-reactivity with other cephalosporins.40
DATs and eluates
DATs were performed in our laboratory on RBCs from 19 of the 25 patients. RBCs
from all 19 patients (100%) reacted with our anti-C3 reagent; four were weakly
reactive (1/2-11/2+), twowere moderately reactive (2-21/2+), and 13 were strongly
reactive (3-4+). RBCs from nine patients (47%) also reacted with anti-IgG; five
were weakly reactive (microscopic-11/2+), one was moderately reactive (21/2+),
and three were strongly reactive (3-31/2+). RBCs from 18 patients were tested with
anti IgM and anti-IgA; four (22%) were reactive with anti-IgM (microscopic to 2+)
and one (6%) was reactive with anti-IgA (1+). Data on four patients who had DATs
performed on multiple samples are shown in Table 2; RBC-bound complement was
detected before RBC-bound IgG in three of these patients and complement was
the only globulin detected on the fourth patients RBCs. Eluates were prepared
from RBCs from 5 of the 25 patients; four of these five patients had RBC-bound
IgG detected (1-3+). All five eluates were nonreactive (two eluates were tested
against drug-treated RBCs, four eluates were tested with reagent RBCs in the
presence of a solution of ceftriaxone, and one eluate was tested at a 1-in-5 dilution
due to limited sample volume).

Drug-treated RBCs
Sera from seven of the 25 patients were tested against drug-treated RBCs (two
ser were tested against RBCs treated using the carbodiimide method, two sera
were tested against RBCs incubated with a ceftriaxone solution prepared in sodium
barbital buffer, and six sera were tested against RBCs incubated with a ceftriaxone
solution prepared in PBS); all either were nonreactive (n = 4) or gave equivalent
reactivity to that seen with untreated RBCs (n = 3). Reactivity with untreated RBCs
was most likely due to the presence of circulating drug or drugimmune complexes
in the sample being tested (see later results and discussion).
Untreated RBCs in the presence of ceftriaxone solution Sera from all 25 patients
were tested in the presence of a 1 mg/mL solution of ceftriaxone. In a few cases,
due to limited sample, for example, from young children, only enzyme-treated
RBCs were tested and/or only tests with fresh normal sera added (as a source of
complement)were performed. In one case, the only sample available from the
patient was plasma, so tests with added fresh normal sera could not be performed.
All 25 patients had ceftriaxone antibodies detected when testing in the presence of

ceftriaxone; Table 3 summarizes the results. Titers that were performed with ser
plus ceftriaxone plus untreated RBCs ranged from 1 to 256 (n = 18; median, 16-32)
for agglutination and from 1 to 8 (n = 9; median, 2) for the antiglobulin test. Titers
that were performed with sera plus ceftriaxone plus enzyme-treated RBCs ranged
from 16 to 1024 (n = 19; median, 128) for agglutination and from 1 to 64 (n = 13;
median, 8) for the antiglobulin test. Serum from only one patient caused in vitro
hemolysis of untreated RBCs in the presence of ceftriaxone (hemolysin titer, 4),
while sera from 16 patients hemolyzed ficintreated RBCs in the presence of
ceftriaxone (hemolysin titers ranged from4 to128;n = 12; median, 16). Intests with
four samples, hemolysis of ficin-treated RBCs was only seen when pooled fresh
normal sera was added to the test system as a source of complement. Three of
these four samples were tested several weeks after being drawn, and thus may
have had depleted complement levels. One patients serum only reacted in the
presence of ceftriaxone when enzyme-treated RBCs were tested (untreatedRBCs
plus ceftriaxonewere nonreactive).Three patients had samples drawn one or more
months after receiving ceftriaxone; two of these patients samples clearly
demonstrated ceftriaxone antibody (strong reactivity vs. untreated RBCs in the
presence of ceftriaxone) when testing sera drawn 4 to 5 weeks after receiving
ceftriaxone. The third patient had several samples tested; sera drawn at the time of
the hemolytic reaction were strongly reactive when testing both untreated and
enzyme-treated RBCs in the presence of ceftriaxone, while a sample drawn 3 to 4
months later was weakly reactive only in tests with enzyme-treated RBCs plus
ceftriaxone (tests with untreated RBCs plus ceftriaxone were negative). Two
patients had samples drawn just before current receipt of ceftriaxone; both
samples demonstrated existing ceftriaxone antibody.
Untreated RBCs with no ceftriaxone added
Sixteen (64%) of 25 patients sera reacted when no ceftriaxone was added, that is,
when PBS was added instead of drug as a dilution control (Table 3). This reactivity
was seen more often when testing enzyme-treated RBCs (n = 16 patients) than
when testing untreated RBCs (n = 2 patients). One sample reacted by antiglobulin

test alone, five samples demonstrated both agglutination and reactivity by the
antiglobulin test, and 10 samples showed agglutination only. Half of the samples
were weakly reactive (1/2-11/2+) and the other half were stronger (2-4+). Fourteen
of the 16 sampleswere tested at a 1-in-10 dilution against enzyme-treated RBCs
and added PBS (no drug added); all 14 were nonreactive (thus titers were less tan
10). One possible explanation for reactivity seen in tests without the in vitro
addition of drug is the presence of circulating drug or drug-antidrug immune
complexes in the patients blood at the time the sample was drawn. Table 4 shows
the temporal relationship of ceftriaxone administration to sample collection and
reactivity for 12 of the 25 patients (we did not have this information for the other 13
patients). Positive PBS controls were seen when testing samples that were drawn
either the same day the patient received ceftriaxone or 1 day after ceftriaxone was
stopped. Positive PBS controls were not seen when testing the samples that were
drawn 2 or more days after stopping ceftriaxone(note that therewere only a few of
those to test). These limited data support the hypothesis that the presence of in
vivo ceftriaxone or ceftriaxone immune complexes in the blood samples was the
cause of the reactive PBS controls.

Gel method
Eleven patients sera were tested with untreated and ficin-treated RBCs in the
presence of ceftriaxone by the gel method.Two sampleswere tested in parallel with
tube testing; the remaining frozen-thawed serawere tested from1 to 23 years
(median, 6 years) after initial tube testswere performed. Although differences were
noted in results using gel compared to tube, in 9 of 11 cases the conclusion that
the patient had an antibody to ceftriaxone could be reached based on the gel
results. Differences between tubeandgel results included: 1) some agglutination of
RBCs detected by the tube method was not detected by gel (e.g., when titers vs.
Untreated RBCs plus ceftriaxonewere_8); 2) in one case, very weak hemolysis
was not noted by the gel method; and 3) antiglobulin results in the presence of
PBS instead of drugwere sometimes stronger by the gel method (e.g., negative or
1/2+ by the tube method but 2+ by the gel method). The missing agglutination
reactions in tests with ceftriaxone and stronger antiglobulin test results in tests with
PBS contributed to the inability to discern the presence of ceftriaxone antibodies in
two patients samples by the gel test.
Immunoglobulin class
Fourteen patients sera were treated with either 2-mercaptoethanol (n = 1) or DTT
(n = 13) and then tested against untreated RBCs in the presence of ceftriaxone. In
nine cases, the agglutinin was shown to be IgM; two had an additional IgG

component. In the remaining five cases, the agglutinin was nonreactive or only
very weakly reactive in the PBS dilution control so no interpretation could be made
with regard to the immunoglobulin class of the agglutinin; one of these patients
antibodies had an IgG component.
Data on patients with no antibodies to ceftriaxone detected
Fifty-four of the 79 tested patients (68%) had no ceftriaxone antibodies detected
when their sera (or plasma) were tested in the presence of a solution of ceftriaxone
and reagent RBCs. Most of these patients had reasonable histories relating
hemolysis to the receipt of ceftriaxone; 10 (19%) of the 54 patients died. Forty-four
of these 54 patients had a DAT performed in our laboratory; 33 had RBC-bound C3
detected (23 had only C3 detected), three had RBC-bound IgG detected but no C3,
and eight had a negative DAT. Eluates were prepared from five patients RBCs
(three had RBC-bound IgG); all eluates were nonreactive (three eluates were
tested against ceftriaxonetreated RBCs and four eluates were tested with
untreated RBCs in the presence of ceftriaxone). Some variations of testing for
ceftriaxone antibodies in the presence of ceftriaxone were tried with a few patients:
1) reagent RBCs were incubated with different dilutions of patients serum
(undiluted, 1/10, and 1/100) and different concentrations of ceftriaxone (0.4 and
400 mg/mL) to look for prozone; 2) washes for the antiglobulin testwere performed
using PBS containing ceftriaxone (0.02 mg/mL); and 3) patients ser and reagent
RBCs were incubated with an ex vivo source of ceftriaxone (urine supplied by Prof.
A. Salama, Berlin, Germany) instead of a solution prepared from ceftriaxone
powder. Despite these efforts, no ceftriaxone antibodies were detected in these
patients samples. Serum from a previously identified patient with ceftriaxone
antibody reacted strongly in the presence of the ex vivo urine source of ceftriaxone.
In some of the 54 patients, other reasons for hemolysis either were determined, for
example, an antibody to another drug (cefotetan13) or a hemolytic transfusin
reaction, or were likely, for example, autoinmune hemolytic anemia (IgG warm, IgM
warm, or mixed IgG warm plus IgM cold) or a congenital hemolytic anemia. Several

patients had warm autoantibodies detectable, especially when testing enzymetreated RBCs. If equivalent reactivity was seen in the presence of PBS and
ceftriaxone, then the interpretation was that no ceftriaxone antibody was detected.
Adsorptions were performed in one case to remove strong autoantibody and the
patients serum was diluted 1 in 2 in another case to weaken the autoantibody;
ceftriaxone antibodies were not detected in either case. In another case with strong
(4+) agglutination of enzyme-treated RBCs with or without ceftriaxone added, the
serum was dialyzed in PBS in an attempt to remove circulating free drug from the
serum, if it was present. The patient was a 5-year-old with acute lymphocytic
leukemia who died after receiving ceftriaxone. The patients serum(an autopsy
sample) only reacted when testing enzyme-treated RBCs (untreated RBCs with or
without ceftriaxone were nonreactive). The serum was dialyzed at 4C for 24 hours
against three changes of PBS using a dialysis cassette (Slide-A-Lyzer, Pierce,
Rockford, IL). The dialyzed serum still caused strong agglutination (3+) of enzymetreated RBCs with or without ceftriaxone added.Dialysis would be expected to
remove free drug but drug-antibody immune complexes, if present, would have
been too large to remove. Although the possibility of ceftriaxone antibody in this
patients sample could not be entirely excluded, it also could not be proven.
Other results
Stability of drug solution
The stability of the 1 mg/mL solution of ceftriaxone in PBS was studied by testing
doubling dilutions of one patients serum (ceftriaxone antibody titers, 64 and 2 for
agglutination and antiglobulin test, respectively) against untreated RBCs in the
presence of ceftriaxone solutions that were freshly prepared, 2 hours old, and 5
hours old. Selected dilutions of another patients plasma (ceftriaxone antibody
agglutinin titer, 256) were tested against untreated RBCs using ceftriaxone
solutions prepared 5 and 3 days previously (stored at 4C) versus one prepared
the day of testing. There were no significant differences in any of the results (data
not shown); thus 1 mg/mL solutions of ceftriaxone in PBS appear to be stable up to
5 days when stored at 4C.

Titrations in fresh normal sera

Eleven patients with ceftriaxone antibodies had dilutions of sera tested to
determine hemolysin titers (dilutions were prepared in fresh normal sera, i.e.,
complement source) in addition to the tests performed to determine agglutination
or antiglobulin test titers (dilutions in PBS). Patients sera diluted in fresh normal
sera to determine hemolysin titers were also examined for agglutination and
indirect antiglobulin test results. It was noted that in tests with enzyme-treated
RBCs, 10 of the 11 patients (91%) had significantly higher agglutination titers
(greater than or equal to a two tube difference) and 8 of the 11 patients (73%) had
significantly higher antiglobulin test titers (greater than or equal to a three tube
difference) when dilutions of patients sera were prepared in fresh normal sera than
when dilutions were prepared in PBS. An example of the results seen with serum
from one patient is given in Table 5. A sample from another patient with ceftriaxone
antibody was tested to determine if preparing dilutions in fresh normal sera was
different from adding fresh normal sera to dilutions prepared in PBS; agglutination
titers were similar when testing enzyme-treated RBCs with sera diluted in PBS
without (titer, <2) or with added fresh normal sera (titer, <2) but were significantly
higher when testing dilutions prepared in fresh normal sera (titer, 32); there was no
significant difference in antiglobulin test results.

It is understandable why the presence of fresh normal sera might enhance the
reactivity of the antiglobulin test (due to increased reactivity by anti-C3) but not why

it would enhance agglutination. One in 100 dilutions of one patients serum (the
patient whose results are given in Table 5) were prepared in different media;
agglutination results of enzyme-treated RBCs in the presence of ceftriaxone were 0
(serum diluted in PBS), 31/2+ (serum diluted in fresh normal sera), 1+ (serum
diluted in pooled normal plasma), and 2+ (serum diluted in 6% bovine albumin). To
determine if this phenomenon would also occur with untreated RBCs plus
ceftriaxone, dilutions of one patients serum were prepared in PBS (agglutination
titer, 64), fresh normal sera (agglutination titer, 8), or 6% albumin (agglutination
titer, 32); titer results of this patients serum with enzyme-treated RBCs plus
ceftriaxone were 512 and 8 (agglutination and antiglobulin test) when serum was
diluted in PBS versus 2048 and 256 when serum was diluted in fresh normal sera.
Thus, this phenomenon appears to occur primarily with enzyme-treated RBCs. For
one patient, the enhanced agglutination seen when testing enzyme-treated RBCs
in the presence of fresh normal sera was important in determining the presence of
a ceftriaxone antibody. The patient was taken off ceftriaxone the same day that the
sample was drawn; thus the presence of circulating drug or immune complexes
probably resulted in positive antiglobulin tests in the PBS controls (untreated and
enzyme-treated RBCs). There was no agglutination noted except when the
patients serum was tested in the presence of fresh normal sera and ceftriaxone
versus enzyme-treated RBCs (3+). Although we add fresh normal sera as a source
of complement (for detecting hemolysis or sensitization with complement by the
antiglobulin test) and we do not understand why fresh normal sera would enhance
agglutination of enzymetreated RBCs by ceftriaxone antibodies, this unexpected
aspect may be helpful in antibody detection in some cases.
DISCUSSION
Ceftriaxone is used to prevent or treat infections in adults and children. Due to its
relatively long plasma half-life, it only needs to be administered once or twice daily,
which makes it a popular cephalosporin for use in pediatric and adult patients. We
have encountered 25 cases of DIIHA caused by ceftriaxone antibodies since 1987.
The majority of these cases (64%) occurred in the past decade, likely due to

increased awareness on the part of health care professionals. The majority (68%)
of patients were children. There were nine deaths (36%) and six of these deaths
were in children. The hemolytic anemia seen in children could be quite dramatic;
symptoms sometimes became apparent within minutes of receiving ceftriaxone
and hemoglobin (Hb) levels dropped very low. The patients had usually received
ceftriaxone on previous occasions without obvious reactions. Ceftriaxone
antibodies fit the category of drug antibodies that are referred to, in the literature,
as reacting by the immune complex mechanism or nonpenicillintype mechanism
or against a neoantigen.30 Typically drug-coated RBCs cannot be prepared in
vitro to test for these types of antibodies. Hemolytic anemia is caused by
complement-mediated reactions; it is often acute, severe, and associated with
hemoglobinemia, hemoglobinuria, and renal failure (i.e., intravascular lysis).
Patients need only take a small amount of the drug. Once the drug is stopped,
hematologic remission is usually rapid.30 Serologically the patients RBCs are
often sensitized with only complement, but IgG is sometimes also present. The
patients serum reacts with RBCs in the presence of the drug and/or a metabolite
of the drug. The IgG and/or IgM serum antibodies almost always activate
complement and cause in vitro hemolysis, agglutination, and sensitization of test
RBCs in the presence of drug; enzyme-treated RBCs almost always react more
strongly than untreated RBCs.30 A patient with DIIHA would be expected to have a
positive DAT at the time of the hemolytic event. The patients in this series with
DIIHA caused by ceftriaxone always had RBC-bound complement detected;
RBCbound IgG was also present about half of the time. DAT results may vary
depending upon when the samples were drawn. Four patients in this study had
DATs performed on multiple samples (Table 2); RBC-bound C3 either was the only
globulin detected or was detected earlier than RBCbound IgG. If a patient was
transfused and/or the simple being tested was drawn several weeks after the
hemolytic event, DAT results could be weak or even negative.41 The few eluates
prepared from patients RBCs were all nonreactive (no ceftriaxone antibody
detected). Although it is unclear why this should be, we have noted nonreactive
eluates (no drug antibody detected) in association with other cases of DIIHA due to

drug antibodies that only react with reagent RBCs in the presence of a solution of
the drug.42-44 In the literature, there are two reports of patients with ceftriaxone
antibodies detected in eluates18,28 and five reports of six patients with nonreactive
eluates3,7,10,11,16 (two of these latter reports are from our group). Ceftriaxone
antibodies were not detected when testing sera against ceftriaxone-treated RBCs,
even when a chemical coupling method (carbodiimide)35,45 was used to try and
attach ceftriaxone to RBCs. Thus, even though most other b-lactam antibiotics that
have been associated with DIIHA (e.g., penicillin, cefotetan) can be shown to bind
covalently to RBC membranes,30,46 we and others have been unable to
demonstrate significant RBC-bound ceftriaxone.47,48 Ceftriaxone antibodies were
only detected when incubating serum and either untreated or enzymetreated RBCs
in the presence of a solution of ceftriaxone.
Detection of hemolysis, agglutination, and sensitization was enhanced by the use
of enzyme-treated RBCs. To detect hemolysis or sensitization by complement, the
patients sample must be serum (not EDTA plasma). The addition of fresh normal
sera to the patients serum was occasionally important for the detection of
hemolysis, usually when the patients sample had been stored for several weeks
before testing. Preparing dilutions of patients sera in fresh normal sera increased
the agglutination and antiglobulin test titers of ceftriaxone antibodies when testing
enzyme-treated RBCs.
Ex vivo ceftriaxone (e.g., urine from a person receiving ceftriaxone) was only used
when testing sera from three patients who did not have ceftriaxone antibodies
detected. It should be noted that sera from the three patients reported in the
literature to react only in the presence of ex vivo ceftriaxone were all tested by the
gel method apparently using untreated RBCs.8,10,12 It would have been of
interest to test these patients sera in the presence of ceftriaxone using enzymetreated RBCs (by the gel or tube methods). There were two samples in our study
that reacted in the presence of ceftriaxone only when using enzyme-treated
RBCs.Our studies with the gel method indicated that direct agglutination of
untreated RBCs plus ceftriaxone may be missed, especially when the agglutinin
titer is 8 or less. We wonder why an antibody to ceftriaxone would react only with

ex vivo ceftriaxone (in urine) when ceftriaxone has been reported to be unchanged
in urine.49 When testing for ceftriaxone antibodies, more tan half of the 25 patients
sera reacted (direct agglutination and/or positive antiglobulin test) even though
ceftriaxone had not been added to the test system in the laboratory. There are
several possible explanations for this scenario: 1) the presence of alloantibodies,
2) the presence of autoantibodies, 3) the presence of circulating drug or drugimmune complexes in the patients blood at the time that the blood sample was
drawn,30,43 or 4) the presence of drug in the test RBC diluent.50 Alloantibodies
had either been detected or ruled out in all of these cases and ceftriaxone is not
used as an antibacterial agent in comercial reagent RBCs so those two
explanations could be excluded. There are a couple of approaches that have been
used to differentiate the presence of autoantibody versus circulating drug or drugimmune complexes in this situation: 1) dialyze the patients serum to remove
circulating

drug,51

2)

use

method

such

as

high-performance

liquid

chromatography to demonstrate the presence of drug in the patients sample,52 or

3) test samples drawn after the drug has had time to clear from the patients
circulation (the time needed depends on the half-life of the drug). The data inTable
4 reflect the last approach. The elimination half-life of ceftriaxone in healthy
persons is 5.8 to 8.7 hours, but in patients with renal impairment (e.g., patients with
intravascular hemolysis caused by a drug antibody) or hepatic dysfunction it can be
longer, for example, 17 to more than 50 hours.53We found that two of six sera
drawn a day after the drug was discontinued and the few examples we had of sera
that were drawn two or more days after the drug was stopped did not react unless
ceftriaxone was added in vitro. This scenario has been reported with a number of
other drugs.46 Since blood samples for drug antibody testing are often drawn while
the patient is still receiving the drug or shortly after the drug has been discontinued,
laboratories need to be aware that patients with drug antibodies may not present
with negative antibody screens.We recommend testing a blood simple drawn 2
days after ceftriaxone was stopped, which should provide clearer serologic
information. Patients with DIIHA may present with a positive DAT and serum
antibody that reacts with all RBCs, thus appearing to have an autoantibody.54,55

Table 8-8 in the Petz and Garratty book Immune Hemolytic Anemias30 lists two
groups of drugs, both reported to induce RBC drug-independent antibodies
(autoantibodies). Group I drugs, for example, methyldopa and fludarabine, induce
drug-independent antibodies only.56,57 Group II drugs, for example, cefotetan and
tolmetin, appear to induce both drug-independent and drug-dependent antibodies.
39,58 Some of the apparent drug-independent autoantibody reactivity seen in
patients taking Group II drugs may be due to the presence of circulating drug or
drugimmune complexes and not to autoantibody (see Garratty and Arndt46 for
further discussion). A nonreactive eluate and temporal relationship of hemolysis
with receipt of a drug are clues that a drug-dependent antibody, rather than an
autoantibody, may be the source of the positive serology.
In addition to the 25 patients in whom we detected ceftriaxone antibodies, we also
worked up 54 additional cases where there was a clinical suggestion that the
patient might have DIIHA caused by ceftriaxone but no ceftriaxone antibodies were
detected by our test methods.
Although most blood samples we tested were drawn close to the time of the
suspected hemolytic event, two of them were drawn 6 months to 1 year later. Data
on one patient with ceftriaxone antibody indicated that after 3 months time, that
antibody was considerably weaker than it had been closer to the time of the
hemolytic anemia (weak reactivity was seen only with enzyme-treated RBCs).
DIIHA is rare and these patients without ceftriaxone antibodies most likely had
hemolysis due to some other reason. Over the years, we have seen increasing
numbers of samples submitted for ceftriaxone antibody workups but only about
one-third had ceftriaxone antibody detected. Increased awareness of the possibility
of DIIHA due to ceftriaxone has most likely contributed to the increase in the
number of cases we have been asked to investigate, especially after the
presentation of these data as an abstract in 2009.59
In conclusion, we describe the serology of 25 cases of DIIHA due to ceftriaxone. All
tested patients RBCs were coated with complement. Patients sera contained IgM
with or without IgG complement-binding ceftriaxone antibodies that often caused in

vitro hemolysis (detection of hemolysis was enhanced by the use of enzymetreated test RBCs and the addition of pooled fresh normal sera as a source of
complement). In the United States, ceftriaxone appears to be, currently, the second
most common drug to cause DIIHA. This is the largest reported series of DIIHA
cases caused by ceftriaxone.